A rule-based question answering system on relevant documents of Indonesian Quran Translation

This paper presents work in development of a question answering (QA) system by using a combination of two different architectures i.e. the one used relevant documents and another used rule-based method, which those two contribute for answer extraction. Base on previous researches testing result, it could be inferred that each of the methods could be a complement for another method in order to increase system performance. This QA was purposed to gather information from Indonesian Quran Translation. The new architecture was designed to gather relevant documents toward the keywords and be used subsequentially to gather answer candidates by using rule-based method. The initial results indicate that system still restricted with retrieved relevant documents, and caused delivering only 60% correct answers. This achievement is not better than the previous one that used rule-based method only.